CD70 expression determines the therapeutic efficacy of expanded human regulatory T cells

Abstract

Regulatory T cells (Tregs) are critical mediators of immune homeostasis. The co-stimulatory molecule CD27 is a marker of highly suppressive Tregs, although the role of the CD27-CD70 receptor-ligand interaction in Tregs is not clear. Here we show that after prolonged in vitro stimulation, a significant proportion of human Tregs gain stable CD70 expression while losing CD27. The expression of CD70 in expanded Tregs is associated with a profound loss of regulatory function and an unusual ability to provide CD70-directed co-stimulation to TCR-activated conventional T cells. Genetic deletion of CD70 or its blockade prevents Tregs from delivering this co-stimulatory signal, thus maintaining their regulatory activity. High resolution targeted single-cell RNA sequencing of human peripheral blood confirms the presence of CD27^-CD70⁺ Treg cells. These findings have important implications for Treg-based clinical studies where cells are expanded over extended periods in order to achieve sufficient treatment doses.

Fig. 1. CD27 and CD70 define distinct human Treg subpopulations.

a Expression of CD27 and CD70 was analyzed on freshly isolated CD4⁺CD25⁺CD127^−/low Tregs. Left panel: Representative FACS plots from one donor. Figures display percentages. Right panel: n = 19 independent donors. Data were analyzed using a Kruskal–Wallis test with Dunn’s post-test for multiple comparisons and are represented as mean ± SD. b FOXP3 expression was measured within the total Treg population or within different Treg subsets defined by CD27 and CD70 expression. Left panel: Representative FACS plots from one donor. Right panel: FOXP3 fluorescence intensity normalized to the total Treg population in different Treg subsets (n = 16 independent donors). Mean ± SD, Kruskal–Wallis test with Dunn’s post-test for multiple comparisons. c, d: c Dot plots for one representative donor in which expression of CD27 and CD70 was measured at different time points on Tregs stimulated with anti-CD3/anti-CD28 coated beads at a ratio of 1:5 beads to cells in the presence of rhIL-2 (250U/ml). Figures show percentages. d CD27 and CD70 fluorescence intensity was analyzed within the CD27⁺CD70⁻ or the CD27⁻CD70⁺ Treg populations, respectively, at different time points and normalized to levels prior stimulation. Data represented as mean ± SD. (n = 6 independent donors with 1–3 replicates per donor). Data were analyzed using a Friedman test with Dunn’s post-test for multiple comparisons. e–g Expression of CD27 and CD70 was analyzed on freshly isolated ex vivo Tregs or Tregs expanded for 2 or 4 weeks with anti-CD3/anti-CD28 coated beads at a ratio of 1:1 beads to cells in the presence of rhIL-2 (1000 U/ml). e Data show n = 11 independent donors analyzed using a Friedman test with Dunn’s post-test for multiple comparisons. f Representative plots for one donor. Figures show percentages. g Figures display percentage and fluorescence intensity of CD70⁺ cells for one representative donor. ns, non-significant.

Fig. 2. Prolonged expansion reduces the suppressive capacity of Tregs.

CD4⁺CD25⁺CD127^−/lowTregs were cultured in vitro with anti-CD3/anti-CD28 coated beads and rhIL-2 before analysis of a FOXP3 expression (n = 8 donors, Friedman test with Dunn’s post-test for multiple comparisons) and b in vitro Treg suppressive capacity, assessed as suppression of responder T cell (Tresp) proliferation. Left panel: percentage of VPD-stained CD4⁺ or CD8⁺ cells in the absence of Tregs (0:1) and in the presence of Tregs at a 1:1 or 1/8:1 Treg:Tresp ratios shown for one representative donor. Right panel: percentage suppression of CD4⁺ and CD8⁺Tresp calculated based on division index as described in “Methods” section. n = 5 donors (three replicate wells for each donor). Mean ± SEM, Friedman test and Dunn’s post-test for multiple comparisons. c In vitro suppressive capacity of CD27⁺CD70⁻ and CD27⁻CD70⁺ sorted Tregs. Left panel: Data from one representative donor. Right panel: Percentage suppression of CD4⁺ and CD8⁺ Tresp for 4–5 independent donors with each data point showing the mean of three replicate wells for each donor; calculated based on division index. Mean ± SEM, Kruskal–Wallis test with Dunn’s post-test for multiple comparisons. Red: CD27⁺CD70⁻Tregs; Blue: CD27⁻CD70⁺Tregs, both sorted after 2 weeks of expansion. d FOXP3 was analyzed in CD27/CD70 Treg subsets after 2 weeks of expansion. Top panel: histogram displays one representative donor (gray: isotype, red: CD27⁺CD70⁻ Tregs, blue: CD27⁻CD70⁺Tregs). Bottom panel: n = 16 independent donors. Mean ± SD, Mann–Whitney test. e, f DNA methylation pattern of the Treg-Specific Demethylation Region (TSDR) was assessed in CD27⁺CD70⁻ and CD27⁻CD70⁺Treg subsets from matched donors, sorted from freshly isolated and 2 weeks in vitro expanded CD4⁺CD25⁺CD127^−/lowTregs. e Average methylation rate for 9 CpG motifs of the TSDR (n = 6 different donors). Top panel: Each box represents the percentage of methylation for each donor. Bottom panel: each dot denotes a donor. Mean ± SD, one-way ANOVA with Tukey post-test. f Each box represents the percentage of methylation of a single CpG residue for each donor at each time point (Left panel: day 0; right panel: 2 weeks) (n = 3 donors matched for each CD27/CD70 Treg subset). Bar colors designate: red: 0% methylation; yellow: 50% methylation; blue: 100% methylation. ns, non-significant.

Fig. 3. Expanded CD27⁺CD70⁻ Tregs retain suppressive potency.

a Schematic overview of Treg expansion strategies. CD4⁺CD25⁺CD127^−/low Tregs were flow sorted from healthy donor PBMCs and expanded in vitro over two rounds of 7 days stimulated with anti-CD3/anti-CD28 coated beads (at a ratio of 1 bead to 1 cell) and rhIL-2 (1000U/ml) and rested for 2 days with a low concentration of rhIL-2 (250U/ml) and without stimulating beads. Tregs expanded for 2 weeks were flow sorted into CD27⁺CD70⁻ and CD27⁻CD70⁺ subsets. Total Tregs (black), flow sorted CD27⁺CD70⁻ Tregs (red) and flow sorted CD27⁻CD70⁺ Tregs (blue) were then cultured for a further 16 days before analysis. Dot plots from one representative donor are shown. Figures display percentages. b Left panel: FOXP3, CD27 and CD70 expression levels on expanded and then flow sorted CD27⁺CD70⁻ and CD27⁻CD70⁺ Tregs after a further 2 weeks of expansion. n = 16–18 independent donors, analyzed using a Mann–Whitney test. Data are represented as mean ± SD. Right panel: Histogram of FOXP3 expression levels for one representative donor (gray: isotype, red: CD27⁺CD70^- Tregs, blue: CD27^-CD70⁺ Tregs). c, d, c In vitro suppression assay of 4 week-expanded Treg populations in one representative donor at multiple Treg:Tresp ratios. Data shown are mean ± SD. d In vitro suppression assays from 14 independent donors at 1:1 (top) and 1/8:1 (bottom) ratios of Treg to Tresp. Each data point represents the mean of three replicate wells from each donor. Mean ± SEM is depicted. Black: total Treg, red: CD27⁺CD70⁻, blue: CD27⁻CD70⁺ Tregs. Data were analyzed using Kruskal–Wallis test with Dunn’s post-test. e The in vivo suppressive capacity of CD27/CD70 Treg subsets was assessed by transplanting VPD-labeled PBMCs together with 4-week expanded CD27⁺CD70⁻ or CD27⁻CD70⁺ Tregs (5:1 PBMC:Treg ratio) into the peritoneal cavity of immunodeficient mice. Top: Schematic representation of the experimental plan. Bottom: The proliferation index of PBMCs was studied by flow cytometry. Data were analyzed using a Kruskal–Wallis test with Dunn’s post-test for multiple comparisons and are represented as mean ± SD. A representative experiment of 1 cell donor out of 2 (4–5 mice per group per experiment, each mouse is represented as a data point). ns, non-significant.

Fig. 4. Transcriptome differences between CD27⁺CD70⁻ and CD27⁻CD70⁺ expanded Tregs.

Thirty-two days-expanded Tregs (as shown in Fig. 3a) were re-sorted into CD27⁺CD70⁻ and CD27⁻CD70⁺ Tregs prior being assessed by RNA-seq analysis. a Heatmap and hierarchical clustering from RNA-seq data with Log2FC ≥ 6. The top 64 differentially expressed genes are shown. Data were analyzed using the edgeR package. b Venn diagram showing the overlap between genes overexpressed in Tregs and upregulated genes in CD27⁺CD70⁻ Tregs or CD27^-CD70⁺ Tregs. c Venn diagram showing the overlap between genes overexpressed in Tconv and upregulated genes in CD27⁺CD70^- Tregs or CD27⁻CD70⁺ Tregs. d, e: d HELIOS expression was measured in freshly isolated total Tregs or within different Treg subsets defined by CD27 and CD70 expression. (Left panel) Representative FACS plots of one donor (gray: isotype, black: total Tregs, red: CD27⁺CD70⁻ Tregs, blue: CD27⁻CD70⁺ Tregs, green: CD27⁺CD70⁺ Tregs, brown: CD27⁻CD70⁻ Tregs). (Right panel) Percentage of HELIOS⁺ cells in different CD27/CD70 Treg subsets. e HELIOS expression was analyzed in 2 week-expanded total Tregs or within different Treg subsets defined by CD27 and CD70 expression. Left panel: CD27/CD70 gating strategy. Figures show percentages. Middle panel: Representative FACS plots of one donor (gray: isotype, black: total Tregs, red: CD27⁺CD70^- Tregs, blue: CD27⁻CD70⁺ Tregs, green: CD27⁺CD70⁺ Tregs, brown: CD27⁻CD70⁻ Tregs). Right panel: Percentage of HELIOS⁺ cells in different CD27/CD70 Treg subsets. Mean ± SD is represented. n = 3–4 independent donors. Kruskal–Wallis test with Dunn’s post-test for multiple comparisons was conducted. Only significant statistical differences are shown. f CD25, GITR, CD39, TIGIT, CD40L, and HLA-DR protein expression was analyzed in 4 weeks expanded CD27⁺CD70⁻ Tregs or CD27^-CD70⁺ Tregs. N = 6–13 independent donors. Data were analyzed by Mann–Whitney test and represented as mean ± SD. ns, non-significant.

Fig. 5. The CD27/CD70 co-stimulatory pathway contributes to Treg function.

Data shown for 4 week-expanded Tregs (sorted as total Tregs, expanded for 2 weeks, sorted by CD27/CD70, then expanded for 2 weeks). a, b: a Cartoon depicting Treg CD27 targeting strategies. Tregs were pre-incubated for 1 h with Fc-mutated blocking anti-CD27 mAb or non-mutated anti-CD27 mAb, followed by 1 h cross-linking with anti-human Ig antibody, or isotype matched antibodies. b Percentage suppression of CD4⁺ and CD8⁺ T cell proliferation by Tregs pre-incubated with anti-CD27 mAbs in 4 donors. CD8⁺T cells did not proliferate enough to perform analysis in 2 out of 4 donors. Friedman test with Dunn’s post-test for multiple comparisons. c Top panel: percentage suppression of CD8⁺Tresp cell proliferation by Treg subsets pre-incubated with anti-CD70 mAb or isotype antibody. n = 5–8 independent cell donors, Wilcoxon matched-pairs test. Bottom panel: proliferation of VPD-labeled CD8⁺Tresp co-cultured with Tregs pre-incubated with isotype or blocking anti-CD70 mAb, one representative donor. d, e: d Cartoon depicting CD27/CD70 blocking strategies. Tregs were pre-incubated with anti-CD70 blocking mAb or matched isotype Ab. Alternatively, soluble Fc-mutated blocking anti-CD27 mAb was added to the suppression assay to block CD27 on Tresp. e Left panel: Percentage of dividing VPD-stained CD4⁺ or CD8⁺ cells in the absence of Tregs (0:1) and in the presence of CD27^-CD70⁺ Tregs pre-incubated with isotype or anti-CD70 blocking mAb (1:1 Treg:Tresp). Right panel: Percentage suppression of proliferation of CD4⁺ and CD8⁺ Tresp. n = 7 independent donors, Friedman test with Dunn’s post-test. b, c, e Graphs calculated based on division index and formula described in “Methods” section. f–h: f VPD-labeled Tresp were co-cultured with CD27⁺CD70⁻Tregs or CD27⁻CD70⁺Tregs pre-incubated with isotype or anti-CD70 mAb in wells containing plate-bound anti-CD3 mAb (1 µg/ml). g Proliferation of CD8⁺ or CD4⁺Tresp in the absence of Tregs or when co-cultured with Treg subsets for one representative donor out of 5. h Division index of proliferating CD4⁺ and CD8⁺Tresp co-cultured with CD27⁻CD70⁺Tregs pre-incubated with isotype mAb or blocking anti-CD70 mAb. n = 7 donors, Friedman test with Dunn’s post-test. b, c, e, h Each data point represents the mean of three replicate wells for each independent cell donor. ns, non-significant.

Fig. 6. CD70 knockout Tregs display enhanced suppressive potency.

a Genetic editing efficiency across the CD70 locus was confirmed by analysis using ICE algorithm. Top panel: Level of discordance between control (orange) and edited (green) genetic sequences. Sequences between control and edited samples match before the PAM sequence (denoted by a dotted line) and are discordant after the cut site. Bottom panel: Graph shows the distribution of the detected indels in the edited sample. X-axis depict the size of the indel and Y-axis shows the frequency of each deletion in the cell product. 1 sgRNA for a representative donor is shown. b Percentage of CD70 and CD27 expression in control and CD70-KO cell products. Assays were repeated in 12 independent experiments. Data were analyzed using a Wilcoxon matched-pairs test. c Treg suppressive capacity was assessed in vitro using autologous VPD-labeled CD3⁺CD25⁻ T cells as responders (Tresp) and allogenic mo-DCs as stimulators. Suppressive activity of control (non-edited) and CD70-KO cell products is represented for six independent experiments. Percentage suppression of proliferation of CD4⁺ Tresp and CD8⁺ Tresp is shown as mean ± SEM. Statistical significance was calculated using a Wilcoxon matched-pairs test. d–f CD70-KO Treg cell products were sorted by magnetic isolation or flow cytometry in CD27⁺CD70⁻, CD27⁻CD70⁺ and CD27⁻CD70⁺ cells subsets before assessing d FOXP3 expression, e DNA methylation pattern of the TSDR, and f suppressive activity. Data shown for 4 independent experiments. d Left panel: histogram displays one representative donor (gray: isotype, red: CD27⁺CD70⁻ Tregs, blue: CD27⁻CD70⁺ Tregs, green: CD27⁻CD70⁻ Tregs). Right panel: Statistical significance was calculated using a Friedman test with Dunn’s post-test for multiple comparisons. e Average methylation rate for nine CpG motifs of the TSDR. Error bars indicate mean ± SD. Statistical significance was determined by a Wilcoxon matched-pairs test. f Treg suppressive capacity was assessed in vitro. Percentage suppression of proliferation of CD4⁺ Tresp and CD8⁺ Tresp is represented. Statistical analyses performed using Friedman test with Dunn’s post-test for multiple comparisons. ns, non-significant.

Fig. 7. Identification of human CD70⁺ Tregs within the peripheral blood by integrated single-cell targeted simultaneous protein and mRNA analysis.

Data from primary resting CD4⁺ cells from the peripheral blood from one systemic lupus erythematosus (SLE) patient, one type I diabetes patient (T1D) and one healthy donor (HD) and analysed using targeted scRNA-seq and protein quantification with BD Rhapsody technology from Trzupek et al. were re-analysed with focus on CD70 expression. A detailed description of this methodology and analysis of these cells can be found in the ref. a Unsupervised hierarchical clustering combining mRNA and protein expression data visualized as a Uniform Manifold Approximation and Projection (UMAP) plot depicting the clustering of memory/effector CD4⁺ T cells from all three donors integrated from two independent experiments. b Heatmap for each CD4⁺ cluster showing the top differentially expressed genes and proteins. c Analysis of CD70⁺ cell frequency within each UMAP cluster in a. The number of cells with ≥1 copy of CD70 is displayed for each cluster. The majority of CD70 mRNA-expressing cells fall within Cluster 0 (Memory Tregs) and Cluster 13 (Activated Tregs). d UMAP plots depicting relative expression of CD70 mRNA for each patient, with Treg clusters highlighted. e Correlation analysis (Spearman correlation coefficient) of CD70 and each assessed mRNA and protein target in the Treg Cluster 0 (top) and Cluster 13 (bottom), with an enhanced view of the top 12 most positively (green) or negatively (red) correlated genes.